Levels of Scientific Evidence

Understanding Evidence-Based Healthcare

“The simple believes everything, but the prudent gives thought to his steps.”
Proverbs 14:15 (ESV)

Introduction

Throughout this book, complementary healthcare interventions have been evaluated according to the principles of evidence-based healthcare. This approach seeks to integrate the best available scientific evidence with clinical expertise, patient values and professional judgement.

Evidence differs in relevance and certainty. A single case report usually cannot estimate treatment effects, while a well-conducted systematic review of suitable studies can provide greater certainty. Study design alone is insufficient: risk of bias, consistency, directness, precision and missing evidence must also be appraised for each important outcome.

This appendix uses a simplified evidence hierarchy as an orientation tool and explains why clinicians must move beyond rank order to question-specific critical appraisal.

Figure B.1

The Evidence Pyramid

Systematic Reviews & Meta-Analyses

▲

Clinical Practice Guidelines

▲

Randomised Controlled Trials

▲

Cohort Studies

▲

Case-Control Studies

▲

Cross-Sectional Studies

▲

Case Series & Case Reports

▲

Expert Opinion & Anecdotal Evidence

B.1 Evidence-Based Healthcare

A foundational definition of evidence-based medicine emphasises:

the conscientious, explicit and judicious use of current best evidence in decisions about individual patient care, integrated with clinical expertise and patient values.

EBHC combines four complementary elements:

• best available scientific evidence;
• professional clinical expertise;
• patient preferences and values;
• clinical circumstances.

Sound decisions integrate relevant evidence, clinical expertise, patient values and circumstances, alongside ethics, resources and applicable professional and legal requirements.

Figure B.2

The Four Pillars of Evidence-Based Healthcare

Scientific Evidence

│

Clinical Expertise

│

Patient Values

│

Clinical Context

│

Wise Clinical Decision

B.2 Why Different Levels of Evidence Exist

The usefulness of a study design depends on the question: intervention effects, diagnosis, prognosis, prevalence, harms and lived experience require different methods.

For example:

• personal testimonies may inspire further investigation but cannot establish effectiveness;
• observational studies can estimate associations and sometimes support causal inference, but require careful control of confounding, selection and measurement bias;
• random allocation reduces confounding on average, but trial validity still depends on allocation concealment, deviations from intervention, missing data, measurement and selective reporting;
• systematic reviews use explicit methods to identify and synthesise eligible evidence, which may be high or low certainty.

Healthcare recommendations should use the best applicable body of evidence and consider certainty for each outcome, benefits and harms, patient values, equity, feasibility and resources.

Table B.1

Hierarchy of Scientific Evidence

B.3 Systematic Reviews and Meta-Analyses

A well-conducted systematic review can provide the most informative synthesis for a defined question, but it is not automatically high-certainty evidence and may miss unpublished, ineligible or inaccessible studies.

When clinically and methodologically appropriate, a meta-analysis statistically combines study estimates and may improve precision. Pooling can also obscure heterogeneity or compound bias, and should not be performed merely because studies report numerical results.

The credibility of a review depends on its question, protocol, search, selection, risk-of-bias assessment, synthesis and included evidence. Multiple biased or indirect studies do not become reliable merely through combination.

Clinical Reflection Box B.1

Reading Beyond the Headline

A media report states that “a herbal supplement is proven to reduce anxiety.”

A prudent clinician asks:

• Was the claim based on a single study or a systematic review?
• Were the included studies methodologically sound?
• Were outcomes clinically meaningful?
• Were adverse effects reported?

Critical appraisal reduces, but cannot eliminate, overinterpretation and should include effect sizes, confidence intervals, absolute effects, multiplicity, funding and certainty of evidence.

B.4 Randomised Controlled Trials

Randomised controlled trials are often the preferred design for estimating causal effects of interventions when ethical and feasible; “gold standard” language can obscure differences in trial conduct and suitability for other questions.

Random assignment can balance prognostic factors on average and support causal inference, provided allocation is concealed and important post-randomisation biases are controlled.

Well-designed RCTs typically include:

• random allocation;
• allocation concealment;
• appropriate control groups;
• blinded outcome assessment where feasible;
• predefined primary outcomes;
• an adequately justified sample size and analysis plan.

RCT limitations may include nonrepresentative recruitment, protocol deviations, attrition, selective reporting, short follow-up and limited power for rare harms. Participants and practitioners may be impossible to blind in complex interventions, but blinded outcome assessment can still be considered.

Table B.2

Strengths and Limitations of Randomised Controlled Trials

B.5 Observational Research

Observational studies examine health outcomes without assigning participants to specific interventions.

Major types include:

• cohort studies;
• case-control studies;
• cross-sectional surveys.

These designs are valuable for prognosis, prevalence, rare or delayed harms and exposures that cannot ethically or practically be randomised. Their contribution depends on design quality and the plausibility of alternative explanations.

Because exposure is not randomly assigned, observational estimates are often more vulnerable to confounding and selection bias. Nevertheless, well-designed observational evidence can sometimes support high confidence, particularly for large effects, dose-response relationships or harms, while a flawed RCT may provide low certainty.

Figure B.3

Observational Studies

Population

│

Observation

│

Association

│

Hypothesis Generation

│

Further Investigation

B.6 Case Reports and Expert Opinion

Case reports can signal unusual events, suspected adverse reactions or hypotheses, but cannot estimate incidence, comparative benefit or causality by themselves.

Nevertheless, isolated clinical experiences are highly susceptible to:

• placebo effects;
• spontaneous recovery;
• regression to the mean;
• observer bias;
• publication bias.

Expert judgement can assist interpretation and decisions under uncertainty, but is vulnerable to conflicts, selective experience and authority bias. Transparent consensus methods should distinguish evidence from judgement and should not substitute for feasible research.

Clinical Practice Box B.2

Interpreting Patient Testimonies

Patient experiences deserve respectful attention and may identify valued outcomes, feasibility or possible benefits and harms. Individual testimonies cannot establish comparative efficacy or frequency and may be influenced by natural history, concurrent care and selective reporting. Decisions should integrate patient goals with the applicable evidence, uncertainty and professional judgement.

B.7 Assessing the Quality of Research

Beyond study design, certainty in a body of evidence is outcome-specific and depends on methodological and statistical factors.

Critical appraisal should evaluate:

• adequacy of sample size;
• randomisation procedures;
• blinding;
• completeness of follow-up;
• validity and reliability of outcome measures;
• statistical analysis;
• conflicts of interest;
• protocol registration, selective reporting, consistency and independent replication.

Reporting guidelines such as CONSORT for randomised trials and PRISMA 2020 for systematic reviews improve completeness and interpretability of reports; adherence does not by itself correct poor design, conduct or analysis.

Table B.3

Checklist for Critical Appraisal

B.8 Applying Evidence to Complementary Healthcare

The principles described in this appendix provide the framework for evaluating complementary therapies throughout this book. No therapy should be accepted solely because it is traditional, widely used or supported by anecdotal success stories. Likewise, potentially beneficial interventions should not be dismissed without careful examination of the available evidence.

Responsible practice integrates rigorous appraisal, clinical expertise, patient preferences, uncertainty, ethics and applicable regulation. It also requires informed consent, monitoring of harms and avoidance of delaying effective care.

Summary of Appendix B

Evidence-based healthcare requires question-specific critical appraisal rather than reliance on isolated experience or a fixed hierarchy. Systematic reviews and well-conducted randomised trials may provide high certainty for intervention effects, while observational designs may be most appropriate for other questions and can sometimes provide compelling evidence. Certainty should be rated for each outcome using risk of bias, inconsistency, indirectness, imprecision and publication bias. Decisions should then integrate benefits, harms, expertise, patient values, equity and context.

Transition to Appendix C – Overview of Complementary Therapies

The next appendix provides a structured overview of selected complementary healthcare modalities discussed throughout this volume, including origins, proposed mechanisms, evidence, potential benefits, risks and Christian considerations. It is an orientation reference, not a substitute for current systematic reviews, clinical guidelines, product-specific safety information or professional judgement.